About
The 1st intelligent insurance documents assessment platform in China that has been validated in large-scale applications, utilizing the state-of-the-art technology of Computer Vision, Natural Language Processing, Cognitive Computing, Knowledge Bases, etc.
Innovation presentation
As a leading insurance company in Asia, Pingan Life serves 64 million policyholders on life and health insurance. During underwriting and claims, we handle massive documents with unstructured data in various formats. Each year, 5 million cases are reviewed manually. This overloaded information greatly affects productivity. statistic shows 65% of underwriters’ time is spent on information extraction. Knowledge workers might also make mistakes. How to avoid repetitive labor and enhance assessment effectiveness is the key to provide faster, easier, and better services. These motivated our platform.
By adopting technology like artificial intelligent, machine learning, and integrating insurance experts’ knowledge, we developed AiDAP, the 1st insurance doc assessment platform in China, validated in large-scale applications. We enhances productivity in three aspects. First, it verifies identities, inspects image quality, and checks completeness to validate submissions. It also automates data entry by extracting information from forms, medical reports and bills to replace human work. Thirdly, we predict health, financial risks and fraud behaviors, and list warnings to knowledge workers.
The platform has served more than 12 million customers, with efficiency increased by 40%, risk avoided over 300 million dollars, resubmission decreased by 80%. Claim processing could be reduced to one minute. Continuity rate also increased by near 1% even during the pandemic.
We believe a platform like AiDAP is meaningful not only because it benefits the carriers. It also helps mitigate losses for individuals and the whole society.
Uniqueness of the project
As a tech-driven platform, AiDAP utilizes the state-of-the-art technology of computer vision, natural language processing, cognitive computing, insurance knowledge bases, etc. It is a multi-modal, cross-disciplinary, intelligent platform with lots of innovations. Besides real-time quality control at the frontend, we automated extraction with joint semantic and visual attributes for higher accuracy and scalability. We also incorporated expert experience, insurance and disease knowledge bases, into machine learning models, and augmented risk identification by visualized anomalies.